This research introduces a systematic framework for identifying fairness issues and intersectional bias in large language models like LLaMA and GPT using metamorphic testing.

• Applies novel fairness-oriented metamorphic relations to assess model bias
• Reveals specific biases affecting multiple demographic intersections
• Provides a structured methodology for bias detection across different LLM architectures
• Emphasizes critical fairness concerns in sensitive applications

From a security perspective, this work enables organizations to identify and mitigate harmful biases before deploying LLMs in high-stakes domains like healthcare, finance, and legal systems, reducing potential discrimination risks and liability.

Metamorphic Testing for Fairness Evaluation in Large Language Models: Identifying Intersectional Bias in LLaMA and GPT